Shows
AI:AMAI:AM — AI Memory Chips, Agent Rankings and AI-Powered Networking · October 28, 2026Kepler Computing CEO Debo Olaosebikan explains how ferroelectric memory could increase cache density and improve high-bandwidth memory using standard fab equipment. Arena CEO Anastasios Angelopoulos discusses agent evaluation, alignment failures and model rankings, while Union CEO Samvit Ramadurgam explores AI-assisted brokering, startup hiring and the continuing importance of human relationships.Chapters(0:00) Memory, rankings and human connections(1:27) Claude Duet gets it done, for a price(3:21) Meet Debo Olaosebikan, Kepler Computing CEO(4:30) Why AI is rationed and where energy goes(6:54) Stack the bits instead of shrinking the transistor
2026-10-291h 16
AI Post TransformersCartridges++: Compressed KV Memory Without Off-Context DerailmentThis episode examines "Cartridges++," an Apple paper on learned, compressed KV-cache memories of a document (Cartridges) and a failure they cause on questions unrelated to that document. It explains how KV-cache compression works and compares three families: eviction methods like H2O and SnapKV, text summaries, and learned methods such as Cartridges, prefix-tuning and Attention Matching. The central argument is that a Cartridge's training objective only rewards matching the teacher on document-related queries and says nothing about unrelated ones. So the memory that best reproduces the document can derail the model elsewhere, for example by quoting a stored policy document...
2026-10-2800 min
MTSWhy AI Agents Deceive Users on Real Tasks | Anastasios AngelopoulosArena CEO Anastasios Angelopoulos discusses Arena's $200M Series B at a $3.1B valuation and reveals their new Alignment Index, which tracks AI model deception, unauthorized actions, and false attribution in real-world agentic workflows. Check out our sponsor Lovable: https://lovable.dev. Turn ideas into software people love.MTS is a live news and interview show covering technology, business, politics, and culture as it happens.Follow MTS: X | https://twitter.com/MTSliveYouTube | https://www.youtube.com/@mtsituationSpotify | https://open.spotif...
2026-10-0924 min
TBPNCrypto Going "Bunker Mode", The Mathpocalypse, Starbucks Explores Chipotle Takeover | Rebecca Kaden & Michael Mignano, Nathan Benaich, Michael Sindicich & Healey Cypher, Moritz Stephan, Anastasios Angelopoulos, Zach Yadegari(01:18) - Crypto Going "Bunker Mode"
(16:31) - The AI Mathpocalypse
(28:08) - Starbucks Explores Chipotle Takeover
(30:57) - Timeline Reactions
(49:01) - Mansion Section
(55:41) - Rebecca Kaden & Michael Mignano discuss USV’s record $900 million fundraise and how the firm is adapting its thesis-driven investment strategy to larger rounds, longer private-company timelines, and AI-enabled opportunities. A partner at USV, Rebecca highlights the firm’s focus on visionary, product-oriented founders, public idea-sharing, emerging technologies, and company incubation.
(01:16:52) - Nathan Benaich discusses the State of AI Report and his balanced, research-driven view of the industry. The Air Street Capital founder highlights AI’s rapid progress, the ri...
2026-10-082h 19MTSCost Per Token Is the Wrong Metric for AI | ArenaArena Co-Founder and CEO Anastasios Angelopoulos breaks down Gemini 4 Argon's performance across text, expert, agent, and coding leaderboards. He discusses why cost-per-task matters more than cost-per-token, style controls for models like Opus 5.5, and benchmarking real-world economic output. Check out our sponsor Lovable: https://lovable.dev. Turn ideas into software people love.--https://x.com/ml_angelopoulos--MTS is a live news and interview show covering technology, business, politics, and culture as it happens.Follow MTS: X | https://twitter.com/MTSliveYouTube | https://www.youtube.com/@mtsituationSpotify...
2026-10-0322 min
Lumeric Daily BriefingLumeric Briefing · 2026-09-11GPT-6 Astra und der Wettbewerb um Coding-Agenten dominieren den Tag – SWE-2, Copilot und Nvidia Sol-Pi zeigen, wie schnell sich Kosten und Fähigkeiten verschieben. Dazu: Destillationsangriffe auf Frontier-Modelle, CPU-Engpässe durch Agenten-Workloads und Googles Datenkauf aus Insolvenzverfahren.
Der Wettbewerb um Coding-Agenten erreicht heute eine neue Intensität. Mit GPT-6 Astra präsentiert OpenAI ein Modell, das Entwickler-Workflows, autonome Computer-Use-Aufgaben und Cybersecurity auf einer Plattform vereint – und als erstes OpenAI-Modell die kritische Cybersecurity-Capability-Stufe des unternehmenseigenen Preparedness Framework erreicht. Astra erzielt 57,9 % auf Terminal-Bench 4.0 und 74,1 % auf DeepSWE v1.1, ist über ChatGPT Plus, Pro, Business und Enterprise sowie die OpenAI API, Microsoft Azure und AWS Bedro...
2026-09-1104 min
GritThe AI Race Has a Leaderboard | Arena CEO Anastasios AngelopoulosAI could be bigger than electricity.Anastasios Angelopoulos is seeing that shift firsthand, with Arena ranking 500+ models and tracking roughly 10 new releases every week.On Grit, he explains why there may never be one dominant AI model and why the API layer may be less sticky than people think.Guest: Anastasios Angelopoulos, Co-Founder and CEO of ArenaConnect with Anastasios AngelopoulosXLinkedIn Connect with MamoonXLinkedInConnect with Joubin:XLinkedInEmail: grit@kleinerperkins.c...
2026-09-071h 00MTSThe Supply Chain Threat to American AI Startups | Anastasios AngelopoulosAnastasios Angelopoulos joins MTS to discuss the latest frontier model releases from Chinese labs, price-performance dynamics on the Arena leaderboards, and the business models behind American open source AI.--https://x.com/ml_angelopoulos--MTS is a live news and interview show covering technology, business, politics, and culture as it happens.Follow MTS: X | https://twitter.com/MTSliveYouTube | https://www.youtube.com/@mtsituationSpotify | https://open.spotify.com/show/4HUNNmV1pjV7CW6Lphm1rM Apple | https://podcasts...
2026-08-0619 min
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch20VC: 70% of Neolabs Will Die | There Will be a $100BN US Open-Source Model | Data is a Trillion $ Market | Governments Cannot Regulate Models: It is Too Late | The Cyber Attacks to Come Will be Insane with Anastasios Angelopoulos @ ArenaAnastasios Angelopoulos is the co-founder and CEO of Arena, the real-world evaluation platform that has become a leading referee of the global AI model race. Arena has raised $250 million, with the latest round valuing the company at $1.7BN. Arena recently surpassed $100M ARR just eight months after launching its enterprise offering, powered by more than 30 million monthly users. AGENDA: 00:00 – Intro: "Kimi beat every American model": What Nobody Wants to Admit… 05:20 – Is this the true commoditization of models? Are they just a utility layer now? 07:30 – Do Chinese open source models cannibalize the closed f...
2026-08-031h 04
This Week in AIWhat happens if China pulls the plug on open-source AI? | Ep 21This Week In Startups is made possible by:PAYPALToday’s show:China is weighing a ban on foreign access to its top AI models… the same low-cost open-weight systems on which half of the Valley’s startups are building. So we asked our two all-star panelists — Arena CEO/Co-Founder Anastasios Angelopoulos and Hippocratic AI CEO/Co-Founder Munjal Shah — what happens to US companies when their favorite cheap, powerful models vaporize overnight.PLUS why Hippocratic runs 31 models in parallel to ensure clinical-grade safety, th...
2026-07-091h 20
AI Post TransformersDiscretizing Reward Models for RL AlignmentThis episode explores the paper Discretizing Reward Models and its argument that smooth decimal reward scores can be misleading for reinforcement learning alignment, because policies learn to exploit tiny, often meaningless differences instead of genuine quality. It explains why reward models are used for fuzzy goals like helpfulness and honesty, then digs into reward hacking, equivalence classes of equally valid answers, and the distinction between a model’s ability to separate good from bad responses versus its tendency to invent rankings among ties. The discussion also covers benchmarks such as the Ties setting and the paper’s core proposal: replacing cont...
2026-07-0600 minMTSClaude Fable 5 Explained in 5 Minutes | 🟡🔴🔵 MTS LiveAnastasios Angelopoulos joins MTS to discuss real-world agentic benchmarks, the breakdown of model performance across tasks, and the challenges of intelligent refusal in frontier AI models.--https://x.com/ml_angelopoulos--MTS is a live news and interview show covering technology, business, politics, and culture as it happens.Follow MTS: X | https://twitter.com/MTSliveYouTube | https://www.youtube.com/@mtsituationSpotify | https://open.spotify.com/show/4HUNNmV1pjV7CW6Lphm1rM Apple | https://podcasts.apple.com/us/podcast/m...
2026-06-1231 min
This Week in AIHow the 1% Will Own Compute (and What It Means for You)The future of AI isn't a smarter chatbot. It's a model that watches your screen, listens to the room, and acts on what it sees. We dug into Thinking Machines' new interaction model, what it means for compute, and the layoff wave that's already here.This week's roundtable: Anastasios Angelopoulos (CEO of Arena, formerly LMArena), Nick Harris (CEO of Lightmatter, photonic computing chips), and Philip Johnston (CEO of StarCloud, building megawatt data centers in space).Thank you to our exclusive sponsor:PayPal Open, One Platform for All Business: http://paypalopen.com/
2026-05-131h 08
This Week in StartupsHow the 1% Will Own Compute (and What It Means for You)The future of AI isn't a smarter chatbot. It's a model that watches your screen, listens to the room, and acts on what it sees. We dug into Thinking Machines' new interaction model, what it means for compute, and the layoff wave that's already here.This week's roundtable: Anastasios Angelopoulos (CEO of Arena, formerly LMArena), Nick Harris (CEO of Lightmatter, photonic computing chips), and Philip Johnston (CEO of StarCloud, building megawatt data centers in space). Thank you to our exclusive sponsor: PayPal Open, One Platform for All Business: http://paypalopen.com/
2026-05-131h 08
EquityThe PhD students who became the judges of the AI industryArtificial intelligence models are multiplying fast, and competition is stiff. With so many players crowding the space, which one will be the best — and who decides that? Arena, formerly LM Arena, has emerged as the de facto public leaderboard for frontier LLMs, influencing funding, launches, and PR cycles. In just seven months, the startup went from a UC Berkeley PhD research project to being valued at $1.7 billion.
On this episode of TechCrunch's Equity podcast, Rebecca Bellan catches up with Arena co-founders Anastasios Angelopoulos and Wei-Lin Chiang to determine how a team like theirs can build a neutral benchmark when the co...
2026-03-1826 min
This Week in AIAI in Warfare, OpenClaw & The Stargate Mega-Campus | This Week in AI E3This week we sit down with three founders building at the frontier of AI infrastructure, evaluation, and hardware: Chase Lochmiller (Crusoe), Naveen Rao (Unconventional AI), and Anastasios Angelopoulos (Arena). We dig into the real bottlenecks slowing down the AI buildout, how to standout in the job market as AI matures, and the explosive government standoff between Anthropic and the Pentagon.We explore how AI infrastructure, open source models, and policy are reshaping the industry from the ground up.The Anthropic-Pentagon Standoff: The U.S. government blacklisted a leading domestic AI company for the first time. Who...
2026-03-041h 16
The Information BottleneckEP26: Measuring Intelligence in the Wild - Arena and the Future of AI EvaluationAnastasios Angelopoulos, Co-Founder and CEO of Arena AI (formerly LMArena), joins us to talk about why static benchmarks are failing, how human preference data actually works under the hood, and what it takes to be the "gold standard" of AI evaluation.Anastasios sits at a fascinating intersection - a theoretical statistician running the platform that every major lab watches when they release a model. We talk about the messiness of AI-generated code slop (yes, he hides Claude's commits too), then dig into the statistical machinery that powers Arena's leaderboards and why getting evaluation r...
2026-02-2444 min
The Information's TITVElon Musk’s “Ego” SpaceX-xAI Bet, Ranking AI Models, Groq’s $7.6 Billion PayoutThe Information’s Editors Martin Peers and Nick Wingfield discuss Elon Musk’s push for orbital data centers and the reality of humanoid robots. We also explore how AI benchmarks can remain fair and unbiased, talking with Arena CEO Anastasios Angelopoulos about the future of AI model ranking. Plus, our AI & Finance Reporter Miles Kruppa discusses Groq’s $7.6 billion payout to shareholders and Blue Owl Capital’s massive data center financier role. Articles discussed on this episode: https://www.theinformation.com/newsletters/dealmaker/groq-shareholders-get-7-6-billion-payouthttps://www.theinformation.com/articles/blue-owl-eyes-new-deals-pushes-deeper-ai-boomSubscrib...
2026-02-1337 min
1st10 Podcast2026 Startup Predictions: Why This Year Changes Everything (Bitcoin, IPOs & The Great Divide)The startup world is splitting into two radically different realities - and 2026 will be the year this divide becomes impossible to ignore. On this episode of the 1st10 Podcast, Boris grades his 2025 forecasts (4.5/5 - not bad!), revealing which bets paid off and where he missed the mark on Bitcoin's meteoric rise. But the real focus is forward: FIVE bold predictions for 2026 that cover everything from crypto's regulatory renaissance and the coming IPO tsunami to a troubling "tale of two cities" emerging between those building the AI-powered future and those struggling to break in. Tune in to hear Boris share...
2026-01-2339 minTBPNDOJ vs. Fed Chair, Apple Repositions Vision Pro, Ben Thompson Joins | Tyler Cowen, Harley Finkelstein, Andrew Feldman, Nathan Nwachuku, Anastasios AngelopoulosSign up for TBPN’s daily newsletter at TBPN.com(00:33) - We Are Jerome Powell
(02:08) - DOJ Opens Criminal Probe into FED Chair
(04:15) - 𝕏 Timeline Reactions
(11:42) - Apple's Vision Pro Strategy
(26:52) - 𝕏 Timeline Reactions
(29:54) - Ben Thompson, founder of Stratechery, is a technology and media analyst known for his in-depth analysis of tech industry strategies. In the conversation, he critiques Apple's Vision Pro, emphasizing that its immersive potential is undermined by overproduced content and unnecessary production elements. Thompson advocates for a more straightforward approach, suggesting that simply placing cameras courtside at live events would pr...
2026-01-133h 21
Latent Space: The AI Engineer PodcastArtificial Analysis: Independent LLM Evals as a Service — with George Cameron and Micah-Hill SmithHappy New Year! You may have noticed that in 2025 we had moved toward YouTube as our primary podcasting platform. As we’ll explain in the next State of Latent Space post, we’ll be doubling down on Substack again and improving the experience for the over 100,000 of you who look out for our emails and website updates! We first mentioned Artificial Analysis in 2024, when it was still a side project in a Sydney basement. They then were one of the few Nat Friedman and Daniel Gross’ AIGrant companies to raise a full seed round from them and have n...
2026-01-081h 18
1st10 PodcastThe AI Leaderboard Every Top Lab Watches - Inside LMArena's Real-World AI BattlegroundWhat happens when three PhD students accidentally build the infrastructure that every major AI lab depends on? OR What if the AI benchmarks everyone trusts are measuring the wrong thing entirely?On this episode of the *1st10 Podcast*, host Boris Epstein sits down with Anastasios Angelopoulos, co-founder and CEO of LMArena, to unpack LMArena went from a Berkeley side project built with free pizza and zero revenue to a $100M company with 42 employees and tens of millions of users in under two years.Tune in to hear them talk about:- How Academic AI benchmarks measure the wrong things...
2026-01-0742 min
Latent Space: The AI Engineer Podcast[State of Evals] LMArena's $1.7B Vision — Anastasios Angelopoulos, LMArenaWe are reupping this episode after LMArena announced their fresh Series A (https://www.theinformation.com/articles/ai-evaluation-startup-lmarena-valued-1-7-billion-new-funding-round?rc=luxwz4), raising $150m at a $1.7B valuation, with $30M annualized consumption revenue (aka $2.5m MRR) after their September evals product launch.—-From building LMArena in a Berkeley basement to raising $100M and becoming the de facto leaderboard for frontier AI, Anastasios Angelopoulos returns to Latent Space to recap 2025 in one of the most influential platforms in AI—trusted by millions of users, every major lab, and the entire industry to answer one question: whic...
2026-01-0624 min
Latent Space: The AI Engineer Podcast[State of Evals] LMArena's $1.7B Vision — Anastasios Angelopoulos, LMArenaWe are reupping this episode after LMArena announced their fresh Series A (https://www.theinformation.com/articles/ai-evaluation-startup-lmarena-valued-1-7-billion-new-funding-round?rc=luxwz4), raising $150m at a $1.7B valuation, with $30M annualized c...Full show notes and links
2026-01-0624 minLatent Space: The AI Engineer Podcast[State of Evals] LMArena's $1.7B Vision — Anastasios Angelopoulos, LMArenaWe are reupping this episode after LMArena announced their fresh Series A (https://www.theinformation.com/articles/ai-evaluation-startup-lmarena-valued-1-7-billion-new-funding-round?rc=luxwz4), raising $150m at a $1.7B valuation, with $30M annualized consumption revenue (aka $2.5m MRR) after their September evals product launch. —- From building LMArena in a Berkeley basement to raising $100M and becoming the de facto leaderboard for frontier AI, Anastasios Angelopoulos returns to Latent Space to recap 2025 in one of the most influential platforms in AI—trusted by millions of users, every major lab, and the enti...
2026-01-0624 minTBPNTrump-Musk Fallout Recap, Circle IPO Post-Game with CEO | Dylan Patel, Jeremy Allaire, Kat Cole, Dana Settle, Anastasios Angelopoulos, Patrick Blumen, Blake Scholl(05:24) - Elon-Trump Crashout Recap
(27:59) - Timeline
(43:34) - Patrick Blumenthal. Patrick discusses his experiences at New York City Tech Week, highlighting the diverse range of industry events and sponsorships. He provides insights into Ukraine's recent military operations, emphasizing the strategic impact of targeting Russian airfields to degrade their nuclear triad and the challenges posed by Russia's porous borders. Blumenthal also shares his perspective on the evolving landscape of drone warfare, noting the rapid advancements and the complexities involved in counter-drone technologies.
(01:00:05) - Jeremy Allaire, CEO and founder of Circle, has a background in building internet software platforms and infrastructure...
2025-06-073h 32
AI + a16zBeyond Leaderboards: LMArena’s Mission to Make AI ReliableLMArena cofounders Anastasios N. Angelopoulos, Wei-Lin Chiang, and Ion Stoica sit down with a16z general partner Anjney Midha to talk about the future of AI evaluation. As benchmarks struggle to keep up with the pace of real-world deployment, LMArena is reframing the problem: what if the best way to test AI models is to put them in front of millions of users and let them vote? The team discusses how Arena evolved from a research side project into a key part of the AI stack, why fresh and subjective data is crucial for reliability, and what it means...
2025-05-301h 41
Best AI papers explainedPrediction-Powered Statistical Inference FrameworkThis paper presents "Prediction-Powered Inference," a novel framework for conducting statistical inference while integrating predictions from machine learning systems with experimental data. The authors, including Anastasios N. Angelopoulos, Stephen Bates, Clara Fannjiang, Michael I. Jordan, and Tijana Zrnic, propose algorithms that calculate provably valid confidence intervals for various statistical measures, like means and regression coefficients, without making assumptions about the specific machine learning model used. A key benefit is that more accurate predictions lead to narrower confidence intervals, increasing data efficiency. The framework's effectiveness is demonstrated through diverse applications, highlighting its potential to improve research conclusions across multiple...
2025-05-0910 min
Neural intel PodGradient Equilibrium in Online LearningGradient Equilibrium in Online Learning is a novel concept introduced in the paper "Gradient Equilibrium in Online Learning: Theory and Applications" by Anastasios N. Angelopoulos, Michael I. Jordan, and Ryan J. Tibshirani. It provides a new perspective on online learning by focusing on the convergence of gradient updates over time, rather than traditional metrics like regret minimization.
2025-02-2015 min
Latent Space: The AI Engineer PodcastIn the Arena: How LMSys changed LLM Benchmarking ForeverApologies for lower audio quality; we lost recordings and had to use backup tracks. Our guests today are Anastasios Angelopoulos and Wei-Lin Chiang, leads of Chatbot Arena, fka LMSYS, the crowdsourced AI evaluation platform developed by the LMSys student club at Berkeley, which became the de facto standard for comparing language models. Arena Elo is often more cited than MMLU scores to many folks, and they have attracted >1,000,000 people to cast votes since its launch, leading top model trainers to cite them over their own formal academic benchmarks: The Limits of Static Benchmarks ...
2024-11-0141 minLatent Space: The AI Engineer PodcastIn the Arena: How LMSys changed LLM Benchmarking ForeverApologies for lower audio quality; we lost recordings and had to use backup tracks. Our guests today are Anastasios Angelopoulos and Wei-Lin Chiang, leads of Chatbot Arena, fka LMSYS, the crowdsourced AI evaluation platform developed by the LMSys stude...Full show notes and links
2024-11-0141 min